Experimental work was divided in two parts. First, the influence of SiN nanomask on the built-in stress in AlGaN/GaN/Si heterostructures was investigated, second, the optimization of growth conditions (mole ratio of V/III group precursors), in order to enhance crystalline quality and electrical properties, was conducted. All samples were grown on 2 inch, 340 nm thick Si(111) substrates in AIXTRON CCS reactor, using trimethylgallium (TMGa), trimethylaluminium (TMAl), ammonia (NH
3) and sillane (SiH
4) chemical reagents as a source of gallium, aluminum, nitrogen and silicon atoms respectively. Total gas flow in the reactor during growth procedure was kept constant at 16 sl/min. Initial thermal cleaning of Si substrates, as well as subsequent growth stages were performed in H
2 atmosphere. In order to prevent formation of parasitic SiN on the Si surface, prior to AlN deposition, the TMAl preflow (20 s, 8.6 µmol/min) without NH
3 was applied. After that, using the same conditions (100 mbar, 1060 °C), NH
3 (67 mmol/min) was introduced and 150 nm thick AlN layer was grown. In the case of samples with SiN nanomask, HT-AlN was in situ covered by the non-continuous SiN layer by switching from TMAl to SiH
4 (0.11 µmol/min) for 100 s. Subsequently, high temperature (Al,Ga)N layers were deposited at 1045 °C and 100 mbar with constant NH
3 (129 mmol/min) flow, whereas low temperature AlN layers at 650 °C with TMAl flow equaled to 9.4 µmol/min. In the first attempt, the impact of non-continous SiN subnanometer layer on the properties of subsequently grown AlGaN/GaN heterostructure was checked. The detailed layers structure scheme and growth parameters of the samples with and without SiN is shown in Table
1 (sample A and B). Application of this layer significantly reduce the built in stress in AlGaN/GaN heterostructure and enhance its crystalline quality (detailed results will be presented later in this manuscript). It led to following investigation of heterostructures containing SiN nanomask. Certain disadvantage of this approach is the increase of the surface defect density—so called micropits [
17], however, by altering of NH
3 and TMGa mole ratio during the growth of GaN buffer, this issue can be effectively solved. High V/III ratios tend to enhance the lateral growth thus suppress formation of these defects on the SiN/GaN interface and in result their reveal on the samples surface. However, in this work, the focus is placed on the influence of V/III growth ratio on the mobility of 2DEG what is an indirect measure of material quality. The layer structure schemes and growth parameters of investigated samples are presented in Table
1 (samples B–E). Different V/III growth ratios were obtained by altering TMGa flow, keeping NH
3 flow constant. V/III mole ratios of 600, 1200, 3000 and 5000 correspond to TMGa flow of 217, 109, 43.5 and 21.7 µmol/min respectivelly at NH
3 flow of 129 mmol/min.
Table 1
The layer scheme and growth parameters of investigated AlGaN/GaN heterostructures with and without SiN mask (A, B) growth with different V/III mole ratios (B–E)
HT-AlN thickness (nm) | 150 | 150 | 150 | 150 | 150 | 150 |
SiN deposition time (s) | – | 100 | 100 | 100 | 100 | 100 |
HT-GaN I thickness (nm); V/III ratio (mol/mol) | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 |
SiN deposition time (s) | – | 100 | 100 | 100 | 100 | 100 |
HT-GaN I thickness (nm); V/III ratio (mol/mol) | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 | 800; 5000 |
LT-AlN thickness (nm) | 15 | 15 | 15 | 15 | 15 | 15 |
HT-GaN II thickness (nm); V/III ratio (mol/mol) | 1600; 1200 | 1600; 1200 | 1600; 600 | 1600; 3000 | 1600; 5000 | 1600; 1200 |
AlN thickness (nm) | \(\sim\) 1.6 | \(\sim\) 1.6 | \(\sim\) 1.6 | \(\sim\) 1.6 | \(\sim\) 1.6 | \(\sim\) 1.6 |
AlxGa1−xN thickness (nm); composition (x) | 22; 0.21 | 22; 0.21 | 22; 0.21 | 22; 0.21 | 22; 0.21 | 22; 0.28 nd = 2 × 1017 cm−3 |